具有自学习能力的变异蝙蝠优化算法及性能仿真

发布时间：2018-01-06 06:28

本文关键词：具有自学习能力的变异蝙蝠优化算法及性能仿真　出处：《系统仿真学报》2017年02期 　论文类型：期刊论文

【摘要】：针对标准蝙蝠算法进化特点,提出了具有自学习能力和个体变异的蝙蝠优化算法,该算法中全局最优个体具有自学习能力,使全局最优解在小范围内进行自我优化,能够引导算法中个体进行深度搜索;同时算法中每个个体动态的成比例形成变异群,依据贪婪选择机制,保护了优良个体,避免个体退化。通过以上算子的融入来提高算法的优化精度和收敛速度,避免算法早熟。通过对基本标准函数的测试,验证了算法具有寻优能力强,搜索精度高的优点,能有效的跳出局部最优,一定程度上弥补了算法高维弱势的缺陷,对于工程中复杂的优化函数有较大的使用价值。
[Abstract]:According to the standard of bat algorithm evolution characteristics is put forward with the bat algorithm self-learning ability and individual variation, the algorithm of the global best individual has the ability of self-learning, the global optimal solution of self optimization in a small range, which can guide the individual search algorithm in depth; at the same time in the algorithm of each individual is proportional to the formation of dynamic variation the group, according to the greedy selection mechanism, the protection of excellent individuals, avoid the individual degradation. To improve the optimization accuracy and convergence speed of the algorithm through the integration of the above operators, avoids premature convergence. Based on the basic standard function tests show that the algorithm has good optimization ability, has the advantages of high search precision, can effectively jump out of local optimal, a certain extent make up for the defects of algorithm of high dimensional weak, has a greater value for the optimization of complex function in engineering practice.

【作者单位】：杭州电子科技大学通信工程学院;杭州电子科技大学电子信息学院;
【基金】：浙江省自然科学基金青年基金(LQ13F 010010) 浙江省重点科技创新团队(2013TD03)
【分类号】：TP18
【正文快照】： 引言1在工程技术、经济管理等诸多领域中,由于实际工程问题的复杂性,大量的优化问题很难通过传统的优化方法来解决。近年来,一类基于生物群体性智能行为的智能优化算法,如粒子群算法[1]、蜂群算法[2]、萤火虫算法[3]等,由于不依赖问题的梯度信息,且具有跳出局部极值点的潜在能

【参考文献】